Use Cases & Prompt Examples

This document provides AI conversation examples for common business scenarios to help you use Lovrabet MCP more efficiently.

---

Quick Start

Before using, ensure:

1. MCP server is configured and connected
2. Logged in to Lovrabet platform
3. Understand basic development terms (datasets, fields, APIs, etc.)

---

Scenario 1: Data Exploration & Understanding

When to Use

• Taking over a new project and need to understand data structure
• Confirming field types and constraints
• Finding related datasets

Prompt Examples

List all datasets containing "customer"

View the detailed information of the customer dataset, including all fields

What are the primary key fields and required fields in the customer table?

What is the relationship between the customer and order tables?

What AI Will Do

1. Call search_datasets or list_datasets tools
2. Call get_dataset_detail to get complete field information
3. Analyze relationships between fields
4. Return structured data structure description

---

Scenario 2: List Page Development

When to Use

• Developing data management pages (table display)
• Need pagination, search, and sorting functionality
• Using Ant Design Pro components

Prompt Examples

Help me develop a customer list page:
1. Use get_dataset_detail to get customer dataset details
2. Use generate_sdk_code to generate filter query code
3. Generate Ant Design Table component with search, pagination, and sorting

Field requirements: display name, phone, status, creation time

Generate code for an order list page:
- Support filtering by status
- Support querying by date range
- 20 items per page, sorted by creation time in descending order

What AI Will Do

1. Get dataset details to confirm field names and types
2. Generate SDK code following Filter query specifications
3. Generate Ant Design Table component code
4. Include complete type definitions and error handling

Filter Query Key Specifications

// Correct: Use operators
where: { status: { $eq: 'active' } }

// Wrong: Direct value
where: { status: 'active' }

// Correct: Array format
select: ['id', 'name', 'email']
orderBy: [{ createTime: 'desc' }]

// Wrong: String format
select: 'id,name,email'
orderBy: 'createTime desc'

---

Scenario 3: Form Page Development

When to Use

• Form pages for creating/editing data
• Need form validation
• Enum field dropdown selection

Prompt Examples

Help me develop a new customer form page:
1. Get required fields from customer dataset
2. Generate Ant Design Form component
3. Include form validation rules
4. Generate SDK code for create operation

Generate code for an order edit page:
- Only allow editing specific fields (status, remarks)
- Required field validation
- Prompt and refresh list after successful submission

What AI Will Do

1. Analyze field metadata (required, enum values, etc.)
2. Generate form components matching field types
3. Add form validation rules
4. Generate submit SDK code

---

Scenario 4: Custom SQL Report Development

When to Use

• Complex data statistics queries
• Cross-table join queries
• Monthly/annual reports

5-Step Mandatory Workflow

Important: Creating custom SQL must follow these steps in order. Do not skip any step.

Step 1 -> Step 2 -> Step 3 -> Step 4 -> Step 5
  ↓        ↓        ↓        ↓        ↓
Query   Generate  Validate   Save    Test

Prompt Examples

Help me create a customer monthly statistics SQL, please strictly follow these steps:

1. Use list_sql_queries to query existing SQL, confirm if it already exists
2. Use get_dataset_detail to get customer table structure
3. Write SQL: count new customers by month
4. Use validate_sql_content to validate SQL syntax
5. Use save_or_update_custom_sql to save SQL
6. Use execute_custom_sql to test execution

SQL name: customer_monthly_stats
SQL code: customer_monthly_stats

Create an order amount statistics SQL:
- Group by customer to calculate total order amount
- Only include orders with status "Completed"
- Display customer name, order count, total amount

What AI Will Do

1. Check if SQL already exists
2. Get table structure to confirm field names
3. Generate SQL with correct syntax
4. Validate SQL correctness
5. Save and test execution
6. Generate SDK code

Common SQL Errors

-- Wrong: Field name may be inaccurate
SELECT customerName FROM customer

-- Correct: First get table structure to confirm field names
SELECT customer_name FROM customer

-- Wrong: Using DELETE (not allowed)
DELETE FROM customer WHERE status = 0

-- Correct: Only SELECT queries allowed
SELECT * FROM customer WHERE status = 1

---

Scenario 5: Backend Function Script Development

When to Use

• Data permission filtering (non-admins can only see their own data)
• Data masking (phone numbers, ID numbers)
• Auto-fill fields (creator, creation time)
• Related query enhancement (order list displays user name)

Script Type Descriptions

Type	Trigger Timing	Typical Use Cases
Before	Before API execution	Permission filtering, parameter validation, auto-fill
After	After API execution	Data masking, field calculation, column hiding
ENDPOINT	Independent endpoint	Complex business logic, transaction processing

Prompt Examples

Help me write a Before script for data permission filtering:
Requirement: When non-admin users query orders, they can only see orders they created

Please follow these steps:
1. Use get_dataset_detail to get order dataset details
2. Analyze which fields to use (created_by, user_id, etc.)
3. Write beforeFilter function
4. Ensure using $and to combine original query conditions
5. Add complete top-level comment documentation

Write an After script for phone number masking:
Requirement: When non-admin users view customer list, display middle four digits as ****

Dataset: customer
Field: phone

Write a Before script to auto-fill when creating orders:
- user_id: current user ID
- tenant_code: current tenant code
- create_time: current time

Write an After script for related query enhancement:
Requirement: Order list displays customer name and level
- Use filter to query user table
- Append user info to order data
- Add new column definitions

Key Specifications

// Filter interface Before: Must operate on ytWhere
export default async function beforeFilter(params, context) {
  if (context.userInfo.role !== "admin") {
    params.ytWhere = params.ytWhere || {};
    const originalWhere = { ...params.ytWhere };
    params.ytWhere = {
      $and: [
        originalWhere,
        { created_by: { $eq: context.userInfo.id } }
      ]
    };
  }
  return params;
}

// Regular interface Before: Direct field assignment
export default async function beforeCreate(params, context) {
  params.created_by = context.userInfo.id;
  params.create_time = new Date();
  return params;
}

// After: Operate on tableData
export default async function afterFilter(params, context) {
  params.tableData?.forEach((record) => {
    if (record.phone) {
      record.phone = record.phone.replace(/(\d{3})\d{4}(\d{4})/, "$1****$2");
    }
  });
  return params;
}

Field Analysis Requirements

Before writing scripts, you must analyze dataset fields:

Please analyze the fields of the order dataset:
1. Which fields are used for permission filtering?
2. Which fields are used for related queries?
3. Which fields need masking?
4. What is the type of each field?

---

Scenario 6: API Integration & Type Generation

When to Use

• Need to auto-generate TypeScript types
• Keep code synchronized with dataset definitions
• Unified API call management

Prompt Examples

Help me generate API call code for customer and order datasets:
1. Use get_dataset_detail to get details for both datasets
2. Use generate_sdk_code to generate filter and create operation code
3. Include complete error handling
4. Add dataset and table comments

Generate code to query orders with status "In Progress":
- Use filter operation
- Sort by creation time in descending order
- 20 items per page
- Return only specified fields

SDK Return Value Handling

// Success: Returns data directly
const result = await client.models.customer.create({ name: "Zhang San" });
console.log(result.id); // Direct access

// Failure: Throws exception
try {
  const result = await client.models.customer.create({ name: "" });
} catch (error) {
  if (error instanceof LovrabetError) {
    console.error("Business error:", error.message);
  }
}

---

Scenario 7: Debugging & Troubleshooting

When to Use

• SQL execution failed
• Query returns empty results
• Permission errors

Prompt Examples

Help me debug why this SQL query returns no results:
SQL code: order_stats
Use execute_custom_sql to execute and analyze errors

Check which fields are available for the filter operation on customer dataset:
1. Use get_operation_detail to get filter operation details
2. Analyze request fields and return values

Validate if this SQL content has syntax errors:
SELECT id, customer_name, create_time FROM customer WHERE status = 1

Use validate_sql_content to validate

---

Common Errors & Solutions

Filter Query Errors

Error Symptom	Cause	Solution
No query results	Forgot to use operators	Use { field: { $eq: value } }
Parameter format error	Wrong parameter name	Use select instead of fields
Sorting not working	Wrong sort format	Use orderBy: [{ field: 'desc' }]
Pagination not working	Wrong parameter name	Use currentPage/pageSize

SQL Related Errors

Error Symptom	Cause	Solution
Field doesn't exist	Didn't confirm table structure	First use get_dataset_detail to confirm fields
SQL save failed	Syntax error	Use validate_sql_content to validate first
No results	SQL code incorrect	Use execute_custom_sql to test

Backend Function Errors

Error Symptom	Cause	Solution
Condition not working	Wrong operator for regular interface	Filter interface uses ytWhere, regular interface direct assignment
Query error	Forgot to use await	All database operations must be awaited
Infinite loop	Recursive call to itself	Prohibit scripts from calling current dataset interface

---

Effective Conversation Techniques

1. Explicitly Specify Using MCP Tools

Use get_dataset_detail to get customer details
Use list_sql_queries to find all SQLs
Use generate_sdk_code to generate code for order filter

2. Describe Complex Requirements Step by Step

Help me complete this task:
Step 1: Query existing customer statistics SQL
Step 2: If not exists, create new SQL
Step 3: Generate call code
Step 4: Test execution

3. Provide Context Information

Based on customer dataset (primary key id, required fields name, phone),
create a new customer form with:
- Name (text input)
- Phone (text input, validate format)
- Status (dropdown: enabled/disabled)

4. Request Following Specifications

When generating code, please follow these specifications:
- Use Filter query operators ($eq, $gte, etc.)
- Add dataset and table comments
- Include complete error handling
- Do not use getList, use filter instead

---

Complete Example: Developing a List Page from Scratch

Help me develop a complete customer list management page:

Step 1: Use get_dataset_detail to get customer dataset details
Step 2: Use get_operation_detail to get filter operation details
Step 3: Use generate_sdk_code to generate filter and delete operation code
Step 4: Generate React + Ant Design component code

Page requirements:
- Search: support searching by name and phone
- Pagination: 20 items per page
- Sorting: default by creation time descending
- Columns: name, phone, status, creation time, actions
- Actions: edit and delete buttons
- Follow Filter query specifications
- Add complete comments

AI will automatically:

1. Call MCP tools to get dataset information
2. Generate SDK code following specifications
3. Generate complete React component
4. Include error handling and loading states

---

Get Help

• Tool Reference - Complete MCP tool documentation
• SDK Return Value Guide - SDK call considerations
• GitHub Issues